Alternative Paths to Competitive Advantage: A Fuzzy-Set Analysis of the Origins of Large Firms

Abstract

Scholars have documented the importance of national-level factors for the competitive success of firms on a global scale. These studies typically identify multiple factors that are behind the emergence of large and successful firms in particular national clusters. However, there has been relatively little research identifying whether such factors are all collectively necessary to produce the outcome, or whether only a few of the factors in different combinations might be sufficient to generate the shift in competitive advantage manifested in the market power of large “flagship” firms. In this paper, we study the evolution of one industry across six countries in which the competitive position of national firms changed considerably during our 100-year analysis period. The results of our combined historical and fuzzy-set analyses show that an unequal distribution of resources may lead to alternative causal pathways to competitive advantage of the largest firms.

Keywords:

• Industry evolution
• competitive advantage of large firms
• national clusters
• history
• paper industry
• fuzzy-set analysis

Acknowledgements

All authors contributed equally. Author names are in alphabetical order. The authors would like to thank the discussants and reviewers of the 2009 Academy of Management Conference and Ecis workshop on “Technological Discontinuities and Transitions: Evolution Theory and Inter-disciplinary Crossovers” in May 2008. An earlier version is the winner of the Carolyn Dexter Award at the 2009 Academy of Management Conference, Chicago. The authors are indebted to Mikko Lauerma for research assistance and Kalle Pajunen for helpful comments. The project was financed by the Academy of Finland and TEKES.

Notes

¹ Although statistical cluster analysis allows studying the effects of different configurations on an outcome of interest, it has a number of known limitations (discussed later in the paper). Fuzzy-set analysis is, however, able to overcome these at least partly (see, e.g. Fiss, 2007; Pajunen, 2008).

² Many scholars also emphasize the importance of small firms in innovation networks and cluster competitiveness. This view has its obvious merits, yet we see that in process industries it is both legitimate and constructive to locate cluster competitiveness in the existence of large firms. More discussion in Patel and Pavitt (1997) and Camison-Zornoza et al. (2004).

³ Simultaneously, the period from 1800 to 2000 witnessed a significant technological shift from hand-made to mechanical manufacturing. In Germany, for example, there were over 1000 paper-producing companies in 1847; most of them were small manufacturers who produced hand-made paper (Krawany, 1910).

⁴ The illustrative 2005 paper machine is the one built by UPM-Kymmene in Rauma, Finland (PK8) in 1998.

⁵ WF = uncoated wood free; LWC = light weight coated; MVC = medium weight coated; WFC = coated wood free.

⁶ Information on Swedish companies can be found at: http://www.svar.ra.se/ (Aktiebolagsdatabas—listed company database).

⁷ On Finnish statistics see: http://www.stat.fi/; on Swedish statistics: http://www.scb.se/; on Norwegian statistics: http://www.ssb.no/; on German statistics: http://www.destatis.de/; on French statistics: http://www.statistique-publique.fr/; on British statistics: http://www.statistics.gov.uk/. See also Eurostat (http://epp.eurostat.ec.europa.eu) and OECD (http://www.oecd.org/statsportal/).

⁸ See http://www.ggdc.net/dseries/—furthermore, a collection of Nordic historical national accounts can be found at: http://avos3.nhh.no/forskning/nnb/

⁹ Although statistical cluster analysis allows studying the effects of different configurations of variables on an outcome of interest, it also has its well-known limitations (see, e.g. Fiss, 2007). For example, cluster analysis tends to treat each configuration as a black box insofar as only differences between constellations of variables can be detected (i.e. the analysis does not extend to the contribution of individual elements to the whole or to an understanding of just how these elements combine to achieve the outcome); cluster analysis methods have high reliance on research judgment (e.g. the choice of a stopping rule); and the cluster solutions for configurations are often highly unstable and their interpretation is frequently difficult. By employing fuzzy-set logic, however, we are able to overcome these problems. This is because fuzzy-set logic relies more on qualitative reasoning based on substantive and theoretical knowledge, and its causation is based on necessity and sufficiency.

¹⁰ We conducted our analyses with fsQCA 2.2 software (Ragin et al., 2006). The program can be downloaded from the website: www.fsqca.com

¹¹ The notations with which the causal conditions can be combined, and which are used in the table, are as follows. First, negation of the causal condition can be calculated simply by subtracting its membership in set A from 1, as follows: (membership in set not-A) = 1 – (membership in set A). In this study, causal conditions are denoted by capital letters. Its negation, on the other hand, is denoted by small letters. Second, intersection takes place when two or more sets are combined. This logical and is accomplished by taking the minimum membership score of each case in the sets that are combined. The minimum membership score, in effect, indicates the degree of membership of a case in a combination of sets (Ragin, 2007). In this study, logical and is denoted by “•”. Third, two or more sets can be joined through logical or (the union of the sets). This is calculated as the maximum of the membership scores, and in this study is denoted by “+”.